CN105045233A - Optimum design method for PID (Proportion Integration Differentiation) controller based on time dimension in heat-engine plant thermal system - Google Patents

Abstract

An optimum design method for a PID (Proportion Integration Differentiation) controller based on a time dimension in a heat-engine plant thermal system. Under a condition that the mathematical model of a controlled object is not completely mastered, the transfer function of a PID controller is GPID (s)=Kf(Kp+Ki/s+Kds/(Tds+1)); a first step of setting is to set parameters ranges of Kp, Ti and Kd of the PID controller, and a golden section method is used to determine values of Kp, Ti and Kd; and a second step of setting is to observe system closed-loop output at set parameters of Kp, Ti and Kd of the PID controller in the first step, and performing parameter setting for two or three times using the golden section method to obtain a satisfied control effect. The invention can complete debugging well and fast and can achieve the satisfied control quality.

Description

Based on the Optimization Design of the PID controller of time metric in Power Plant Thermal system

Technical field

The present invention relates to a kind of in Power Plant Thermal system, optimization based on the model-free PID controller of time metric is adjusted method for designing, golden section optimization method is applied in design, with other apply golden section optimization method design PID controller parameter unlike, method for designing of the present invention does not need by the mathematical model of control object, and increases pre-factor K _fdesign, this is the principal feature of method for designing of the present invention.It is applied to Power Plant Thermal Systematical control, belongs to automatic control technology field.

Background technology

In industrial process control, PID controller is most widely used controller.Usually, engineers design's PID controller, needs the accurate model knowing controlled device, but in Practical Project, accurate model is often difficult to obtain.When model-free, design and Tuning PID Controller parameter are complicated debug processs, if the engineering technical personnel at scene do not understand the characteristic of controlled device, unsuitable adjustment method may make system disperse, cause industrial accident, this debug process is a job of wasting time and energy, and uncertainly can obtain satisfied Control platform.Present patent application is invented and is devised design and the setting method of measuring the model-free PID controller of indignant thought based on system time, and it better can complete debug process faster, and can obtain satisfied Control platform.Apply golden section optimization method in program, with other apply golden section optimization method design PID controller parameter unlike, present patent application method for designing does not need by the mathematical model of control object, and increases by pre-factor K _fdesign, this is the principal feature of present patent application method for designing.

Summary of the invention

Controller in modern control system is all that appliance computer realizes, and each control system has certain control cycle (or sampling period) t _s, t _srelevant to the predominant frequency of computing machine.

Definition a: analog control system controls (operation) cycle t at it _swhen determining, after steering order (being generally step signal) sends, system when not exceeding steering order value, the time T=nt of steering order value that quick and stable completes (reaching) _s.Wherein T is system non-overshoot settling time, and n is the time metric of control system.

Design and the setting method of measuring the model-free PID controller of indignant thought based on system time of present patent application invention and design are exactly control cycle (or sampling period) t according to control system _sset out.

PID controller parameter based on time metric is adjusted design procedure

Step one: do not comprise unstable limit and pure delay time in controlled device, in the control system built by PID controller, T integral time of PID controller _i≈ (1/3 ~ 1/6) nt _swherein n is the time metric of control system.If controlled device is G (s), if open cycle system Q (s)=G (s) G after PID compensates _pID(s) ≈ 1/T _is, then closed-loop system H (s)=1/ (T _is+1), this system completes the time T ≈ 6T of steering order (step signal) _i.

If the open cycle system after PID compensates (in most cases, the transport function of open cycle system can be of equal value therewith):

Q(s)＝G(s)G _PID(s)＝1/(cs ²+ds)＝1/(T _is(T _is/4+1))

Due to system non-overshoot, damping coefficientζ=1, then c=1/w ², d=2/w, T _i=d.Closed-loop system H (s)=1/ (cs ²+ ds+1), complete the time T ≈ 7/w of steering order, then T _i≈ (2/7) T.

In formula, symbol description is as follows: T ≈ 7/w, w are system undamped vibration frequencies.

This is the basis of integral time in Patent design PID controller of the present invention.

The transport function of traditional PID controller is: G _pID(s)=K _p+ K _i/ s+K _ds

Step 2: the transport function of the PID controller that the present invention improves is:

G _PID(s)＝K _f(K _p+K _i/s+K _ds/(T _ds+1))(1)

Wherein K _f, T _deffect be very important, select suitable K _fthe effect of getting twice the result with half the effort can be obtained to the PID controller design of Large time delay process system.

K _f=f (τ/t _p)=f (h), h=τ/t _p; Wherein, τ is system pure delay time, t _pwhen without pure delay time τ, the settling time of system, K _fthe monotonic decreasing function about h, 0 < K _f≤ 1, (when τ=0, K _f=1).The present invention obtains formula below by the method for a large amount of project data and curve:

K _f＝f(τ/t _p)

K _f＝f(h)＝c ₀+c ₁h+c ₂h ²+c ₃h ³+c ₄h ⁴+…，(2)

Or $K_f=f\left(h\right)=e^{-(c_0+c_{1}h+c_2{h}^2+c_3{h}^3+c_4{h}^4+...)}---\left(3\right)$

Facts have proved, as coefficient h=τ/t _pduring increase, in order to the stable of Guarantee control system and quality, K must be reduced _f.

Calculating K above _fformula non-linear, compensate for just by control object G (s)=k ₀e ^{-τ s}/ (as ²+ bs+1)=e ^{-τ s}g ₁non-minimum phase bit position e in (s) ^{-τ s}to minimum phase part k/ (as ²+ bs+1) impact.

Differential K in conventional PID controllers _ds is improved to K _ds/ (T _ds+1), effectively can improve the percussive action of differential, pass through T _dsetting can increase time of the differential action, reduce the interference that differential brings.

Step 3: when not exclusively knowing controlled device mathematical model, if set the time scale of control system as n, according to above-mentioned proposition, can according to method Tuning PID Controller parameter below.

In practical engineering application, at scattered control system (DCS) control cycle t _swhen determining, this PID controller parameter setting method does not need the mathematical model knowing controlled device, only needs 1 basic parameter, i.e. time scale n of roughly certainty annuity.The time T that system completes steering order generally also easily obtains (T=nt at the scene _s), which reflects control engineering teacher to the basic understanding of controlled system characteristic and understanding, such as in fuel-burning power plant in the systems such as steam temperature, water level, vapour pressure, load, T is different.

If can not n be estimated, just can only according to the control cycle of system (or sampling period) t _sbe optimized design.

(1) if can estimate n, T integral time is set _i=n _it _s, wherein n _i=(1/3 ~ 1/6) n.If can not estimate n, T integral time is set _i=(60 ~ 200) t _s, initial setting pid parameter is as follows:

PID controller pre-factor K _f=1, Proportional coefficient K _p=(0.3 ~ 1.5); , differential coefficient K _d=0, derivative time T _d=(1 ~ 5) t _s.General, there is large fluctuation in order to anti-locking system exports, need integral time selection larger, get T _i=200t _s, at Kp, T _i, K _dgo control system in this group parameter area, can ensure that controlled volume there will not be large fluctuation.Application golden section approach, can organize in parameter area at this and determine a K _p, T _i, K _dvalue.

(2) apply the pid parameter of above-mentioned first step setting, if controlled device exports slow, application Fibonacci method reduces T integral time _i, increase Proportional coefficient K _pif (can n be estimated, arrange T _i=nt _s/ (4.5 ~) 6; ) or increase K _f, K _f=(1.1 ~ 1.7).

Apply the pid parameter of above-mentioned first step setting, if the output of controlled device is too fast and fluctuation is large, illustrate that controlled device comprises pure delay time link, or the damping coefficientζ < 0.76 of controlled device, can apply Fibonacci method increases T integral time _i, reduce Proportional coefficient K _p, (if can estimate n, T is set _i=nt _s/ (2.5 ~ 3.5)).If the output of controlled device still has fluctuation, but fluctuation reduces to some extent, can add differential K _ds/ (T _ds+1) (K is acted on _d=T _i/ 4, T _d=(1 ~ 5) t _s); If after adding the differential action, controlled device exports still has fluctuation, can reduce K _f, get K _f=0.1 ~ 0.7.

(3) if through (1), the design of (2) step still can not obtain satisfied Control platform, needs to get back to the first step and resets K _p, T _i, K _dspan, repeat (1) and (2) step design until obtain be satisfied with Control platform.

According to above-mentioned (1), the step design tuning PID controller parameter of (2) and (3), this pid parameter setting method based on time scale, in (2) step, changes K _fbe equivalent to and change other 3 parameter K simultaneously _p, T _i, K _d, this is a kind of simple effective method, namely can obtain satisfied Control platform general 2 ~ 4 times.

Advantage and effect: based on the Optimization Design of the PID controller of time metric in a kind of Power Plant Thermal system of the present invention, its advantage measures design and the setting method of the model-free PID controller of indignant thought, it better can complete debug process faster, and can obtain satisfied Control platform.Apply golden section optimization method in program, with other apply golden section optimization method design PID controller parameter unlike, present patent application method for designing does not need by the mathematical model of control object, and increases by pre-factor K _fdesign, this is the principal feature of present patent application method for designing.

Accompanying drawing explanation

Fig. 1 typical PID control system schematic diagram.

The pid parameter control system closed loop output response schematic diagram of Fig. 2 first time design tuning.

The pid parameter control system closed loop output response schematic diagram of Fig. 3 second time design tuning.

The control action schematic diagram of P, I, D in Fig. 4 PID controller.

Fig. 5 a, b are the program interface schematic diagram of design.

Fig. 6 is calculation procedure block scheme.

Embodiment:

The PID controller of adjusting designed by the present invention needs to realize in the upper configuration of the distribution type control system (DCS) of user, then carries out real time execution control.Also can implement on industrial computer.

According to the method for designing in summary of the invention, in conjunction with ratio P in PID controller, the effect (see Figure of description 4) of integration I and differential D, the PID controller parameter program of adjusting of the present invention's design is as shown below:

Apply golden section (0.618) optimization method in program, with other apply golden section optimization method design PID controller parameter unlike, this patent method for designing does not need by the mathematical model of control object, and increases pre-factor K _fdesign, this is the principal feature of method for designing of the present invention.

So-called golden section (0.618) optimization method program is:

function[xo,fo]＝Opt_Golden(f,a,b,TolX,TolFun,k)

%%%% golden search algorithms asks the optimum solution on interval [a, b]

%f is objective function, and TolX is x threshold value, and TolFun is function threshold, and k is iterations

R=(sqrt (5)-1)/2=0.618; %r is golden section point value,

H=b-a; % interval width

rh＝r*h；

%%% gets c, d at 2, and calculates corresponding functional value fc and fd

c＝b-rh；

d＝a+rh；

fc＝feval(f,c)；

fd＝feval(f,d)；

%%% algorithm second step judges whether to stop iteration

%%%% algorithm the 3rd step, carries out new round iteration

Above in program Opt_Golden (f, a, b, TolX, TolFun, k)

Performance index f determines according to the computing method of analysis below

From perturbation analysis, when the disturbance of system is deterministic perturbation, the performance evaluation of its control system cannot directly apply minimum variance index.Usually under deterministic perturbation, the performance of system is divided into dynamic property and steady-state behaviour, is divided into again time domain performance and frequency domain performance from research field.The steady-state behaviour of time domain adopts steady-state error usually.The dynamic performance index of time domain is usually as follows:

Rise time: unit-step response rises to 90% of steady-state value from 10% of steady-state value, the required time.System for system unit step response monotone variation is like this, and to there being the system of concussion, can be defined as the time arriving stable state from initial value to first time.Rise time, reaction be the response speed of system.

Time to peak: for the system having concussion, its unit-step response is passed through final value and is reached first peak value time used.

Regulating time: unit-step response reaches the error band of 5% (2%) of steady-state value and the time of maintenance first.

Overshoot: when the unit step of system has overshoot, the peak-peak of its unit-step response deducts the percentage of the difference of steady-state value and the ratio of steady-state value.

Also a series of index is had to carry out the performance of reactive system in a frequency domain, as reflected the zero-frequency value of systematic steady state precision, the resonance peak of reflection system overshoot situation, the bandwidth of reflection system fast response characteristic and anti-interference filtration characteristic, the phase margin of reflection systematic steady state characteristic and magnitude margin etc.

Error intergal performance index

The various indexs introduced in upper joint are performance index of individual event, and the integrated form some integrated performance indexs being included to error has various forms, and conventional having is following several:

1) error intergal (IE)

$IE={\&Integral;}_0^{\mathrm{\&infin;}}e\left(t\right)dt$

2) absolute error index (IAE)

$IAE={\&Integral;}_0^{\mathrm{\&infin;}}\left|e\left(t\right)\right|dt$

3) integral square error (ISE)

$ISE={\&Integral;}_0^{\mathrm{\&infin;}}{e}^2\left(t\right)dt$

4) time and Error Absolute Value integration (ITAE)

$ITAE={\&Integral;}_0^{\mathrm{\&infin;}}t|e\left(t\right|)dt$

The estimation of above-mentioned various index be situation for whole process but emphasis is had nothing in common with each other.The error gentle (SSE) of more brief and practical in actual performance is evaluated:

$SSE=\frac{1}{n}\underset{i=1}{\overset{n}{\&Sigma;}}{{\mathrm{\&epsiv;}}^2}_i$

Can the determinacy performance index of define system based on this:

${\mathrm{\&eta;}}_d=\frac{\underset{i=0}{\overset{d-1}{\&Sigma;}}{e}^2\left(i\right)}{\underset{i=0}{\overset{\mathrm{\&infin;}}{\&Sigma;}}{e}^2\left(i\right)}$

The Unified Form of performance index

No matter what exist in loop is deterministic perturbation or randomness disturbance, B.Huang is unified H ₂performance index under norm meaning:

$\mathrm{\&eta;}=\frac{min\left(\right|\left|G_{cl}\right|{|}_2^2)}{\left|\right|G_{cl}|{|}_2^2}$

When system only has randomness disturbance, assuming that the setting value of system is 0, namely as randomness disturbance α _t≠ 0, deterministic perturbation time performance evaluation as follows:

Its closed loop H ₂norm is defined as follows:

$\left|\right|G_{cl}|{|}_2^2=\frac{1}{2\mathrm{\&pi;}}{\&Integral;}_{\mathrm{\&pi;}}^{\mathrm{\&pi;}}tr\&lsqb;G_{cl}\left(e^{jw}\right)G_{cl}^T\left(e^{-jw}\right)\&rsqb;dw$

Wherein: G _clfor control system closed loop transfer function,

Randomness disturbance is write as polynomial form to the transport function of departure:

$G_{cl}=\underset{i=0}{\overset{\mathrm{\&infin;}}{\&Sigma;}}{f}_{\mathrm{\&alpha;}}\left(i\right)z^{-i}$

Now the stochastic performance index of system is defined as follows:

${\mathrm{\&eta;}}_s=\frac{min\left(\right|\left|G_{cl}\right|{|}_2^2)}{\left|\right|G_{cl}|{|}_2^2}=\frac{\underset{i=0}{\overset{d-1}{\&Sigma;}}{f}_{\mathrm{\&alpha;}}^2\left(i\right)}{\underset{i=0}{\overset{\mathrm{\&infin;}}{\&Sigma;}}{f}_{\mathrm{\&alpha;}}^2\left(i\right)}=\frac{{\mathrm{\&delta;}}_{mv}^2}{{\mathrm{\&delta;}}_y}$

η _smore close to 1, think that the performance of system is better.The computing method of its stochastic performance index can be drawn from formula right-hand component.Estimation for minimum variance can have been come by FCOR algorithm or ARIMA modeling.

When system only has setting value disturbance, performance evaluation being carried out to it and it is only there being the evaluation in randomness disturbance situation similar, starting with from departure, defining its tracking performance index.Now, α _t=0 formula is easy to get:

Now, its closed loop H ₂norm is defined as follows:

$\left|\right|G|{|}_2^2=G_{SEE}=\frac{1}{n}\underset{i=1}{\overset{n}{\&Sigma;}}{{\mathrm{\&epsiv;}}^2}_i$

Tracking performance (determinacy performance) index now ^[44]be defined as follows:

${\mathrm{\&eta;}}_d=\frac{min\left(\right|\left|G_{cl}\right|{|}_2^2)}{\left|\right|G_{cl}|{|}_2^2}=\frac{\underset{i=0}{\overset{d-1}{\&Sigma;}}{f}_{\mathrm{\&zeta;}}^2\left(i\right)}{\underset{i=0}{\overset{\mathrm{\&infin;}}{\&Sigma;}}{f}_{\mathrm{\&zeta;}}^2\left(i\right)}=\frac{\underset{i=0}{\overset{d-1}{\&Sigma;}}{\mathrm{\&epsiv;}}^2\left(i\right)}{\underset{i=0}{\overset{\mathrm{\&infin;}}{\&Sigma;}}{\mathrm{\&epsiv;}}^2\left(i\right)}$

Easily find out from the definition of this index, step signal on the perfect tracking that system is faster after dead band, then η _dmore close to 1, the tracking performance of system is better.

Other performance index

The Performance Evaluating Indexes of control system, except minimum variance, also has other indexs many.

First based on the index of historical data be easily expect and also practicality.Operations staff is by oneself observation and a certain control system of empirical discovery, control effects within certain a period of time is ideal, so based on the service data in this period, calculate value (as variance) now as the benchmark to this system evaluation according to a certain algorithm.In this can be based on the performance index of historical data:

${\mathrm{\&eta;}}_{his}=\frac{J_{his}}{J_{act}}$

But certainly close to 1, this index more thinks that the performance of system performance is more satisfactory.Its shortcoming is that this desired value is limited between [0,1] unlike minimum variance evaluation index, and its scope is indefinite.But its advantage is also clearly, this index overcomes minimum variance performance index and makes control action too fierce in some cases and the shortcoming that cannot realize, because must reach based on the performance number of historical data, and calculating is easily, and easy to understand.

Secondly, the minimum variance (GeneralisedMinimumVariance, GMV) of broad sense is also a conventional index.Calculate MV index time, calculating be system export y _tor error e _tvariance under control law, and this control law usually causes the control action of system too drastic and cannot realize LMS control in practice.For overcoming this shortcoming, GMV is by controlled quentity controlled variable u _twith error e _tbe weighted structure new variables, control law is by realizing this new variables least squares optimization.By the selection of the weighting coefficient to new variables, consideration is carried out to the control action of system and avoids control procedure too drastic, thus the control law obtained can be used in practice.Designed program interface is as Fig. 5 a, Fig. 5 b, and Fig. 6 is shown in by calculation procedure block scheme.

The present invention's design (increases pre-factor K artificial experience _fwith the estimation to integral time) combine with golden section (0.618) optimization method, devise control system below.

Figure of description 1 is the typical PID control system block diagram designed in Power Plant Thermal process.

Under coal-supplying amount and the indeclinable situation of absorbing quantity, to be the transport function between cigarette vapour oxygen level and air output be controlled device:

G(s)＝1.3e ^-5s/(54s ²+14s+1)；

Oxygen amount set-point is 4.3 (%), control cycle t _s=0.2 (s);

According to calculation procedure above, process of optimization is as follows:

1. first step parameter designing: get K _f=1; t _i=80t _s=16; k _p=0.62; k _d=0.0; Td=1;

Figure of description 2 is application methods for designing of the present invention, the pid parameter control system closed loop output response of first step design tuning.

2. as can be seen from Figure 2, the first time pid parameter output of design tuning has overshoot, second time parameter designing: get K _f=0.55; Other parameter constant, control system closed loop output response as represented in fig. 3.Second time parameter designing value control effects is fine.

Fig. 4 is the control action of P, I, D and PID in PID controller.

Fig. 5 a, b are the program interface schematic diagram of design, and Fig. 6 is calculation procedure block scheme.

Claims (1)

1. in Power Plant Thermal system based on the Optimization Design of the PID controller of time metric, it is characterized in that: the method concrete steps are as follows:

Step one: do not comprise unstable limit and pure delay time in controlled device, in the control system built by PID controller, T integral time of PID controller _i≈ (1/3 ~ 1/6) nt _swherein n is the time metric of control system; If controlled device is G (s), if open cycle system Q (s)=G (s) G after PID compensates _pID(s) ≈ 1/T _is, then closed-loop system H (s)=1/ (T _is+1), this system completes the time T ≈ 6T of steering order and step signal _i;

If the open cycle system after PID compensates is

Q(s)＝G(s)G _PID(s)＝1/(cs ²+ds)＝1/(T _is(T _is/4+1))

Due to system non-overshoot, damping coefficientζ=1, then c=1/w ², d=2/w, T _i=d; Closed-loop system H (s)=1/ (cs ²+ ds+1), complete the time T ≈ 7/w of steering order, then T _i≈ (2/7) T;

In formula, symbol description is as follows: T ≈ 7/w, w are system undamped vibration frequencies;

This is the basis of integral time in design PID controller, and the transport function of traditional PID controller is:

G _PID(s)＝K _p+K _i/s+K _ds；

Step 2: the transport function of the PID controller of improvement is:

G _PID(s)＝K _f(K _p+K _i/s+K _ds/(T _ds+1))(1)

Wherein K _f, T _dimportant role, select suitable K _fvery crucial to the PID controller design of Large time delay process system;

K _f=f (τ/t _p)=f (h), h=τ/t _p; Wherein, τ is system pure delay time, t _pwhen without pure delay time τ, the settling time of system, K _fthe monotonic decreasing function about h, 0 < K _f≤ 1, when τ=0, K _f=1; Formula by the method for a large amount of project data and curve obtains below:

K _f＝f(τ/t _p)

K _f＝f(h)＝c ₀+c ₁h+c ₂h ²+c ₃h ³+c ₄h ⁴+…，(2)

Or $K_f=f\left(h\right)=e^{-(c_0+c_{1}h+c_2{h}^2+c_3{h}^3+c_4{h}^4+...)}---\left(3\right)$

Facts have proved, as coefficient h=τ/t _pduring increase, in order to the stable of Guarantee control system and quality, K must be reduced _f;

Calculating K above _fformula non-linear, compensate for just by control object G (s)=k ₀e ^{-τ s}/ (as ²+ bs+1)=e ^{-τ s}g ₁non-minimum phase bit position e in (s) ^{-τ s}to minimum phase part k/ (as ²+ bs+1) impact;

Differential K in conventional PID controllers _ds is improved to K _ds/ (T _ds+1), effectively can improve the percussive action of differential, pass through T _dsetting can increase time of the differential action, reduce the interference that differential brings;

Step 3: when not exclusively knowing controlled device mathematical model, if set the time scale of control system as n, according to above-mentioned proposition, according to method Tuning PID Controller parameter below;

In practical engineering application, at distributed monitoring control system control cycle t _swhen determining, this PID controller parameter setting method does not need the mathematical model knowing controlled device, only needs 1 basic parameter, i.e. time scale n of roughly certainty annuity; The time T that system completes steering order generally also easily obtains T=nt at the scene _s, which reflects control engineering teacher to the basic understanding of controlled system characteristic and understanding, in fuel-burning power plant in steam temperature, water level, vapour pressure, load system, T is different;

If can not n be estimated, just can only according to the control cycle of system or sampling period t _sbe optimized design;

(1) if can estimate n, T integral time is set _i=n _it _s, wherein n _i=(1/3 ~ 1/6) n; If can not estimate n, T integral time is set _i=(60 ~ 200) t _s, initial setting pid parameter is as follows:

PID controller pre-factor K _f=1, Proportional coefficient K _p=(0.3 ~ 1.5); , differential coefficient K _d=0, derivative time T _d=(1 ~ 5) t _s; General, there is large fluctuation in order to anti-locking system exports, need integral time selection larger, get T _i=200t _s, at Kp, T _i, K _dgo control system in this group parameter area, can ensure that controlled volume there will not be large fluctuation; Application golden section approach, determines a K in this group parameter area _p, T _i, K _dvalue;

(2) apply the pid parameter of above-mentioned first step setting, if controlled device exports slow, application Fibonacci method reduces T integral time _i, increase Proportional coefficient K _pif can n be estimated, arrange T _i=nt _s/ (4.5 ~ 6) or increase K _f, K _f=(1.1 ~ 1.7);

Apply the pid parameter of above-mentioned first step setting, if the output of controlled device is too fast and fluctuation is large, illustrate that controlled device comprises pure delay time link, or the damping coefficientζ < 0.76 of controlled device, application Fibonacci method increases T integral time _i, reduce Proportional coefficient K _pif can n be estimated, arrange T _i=nt _s/ (2.5 ~ 3.5)); If the output of controlled device still has fluctuation, but fluctuation reduces to some extent, adds differential K _ds/ (T _ds+1) K is acted on _d=T _i/ 4, T _d=(1 ~ 5) t _s; If after adding the differential action, controlled device exports still has fluctuation, reduces K _f, get K _f=0.1 ~ 0.7;

(3) if through (1), the design of (2) step still can not obtain satisfied Control platform, needs to get back to the first step and resets K _p, T _i, K _dspan, repeat (1) and (2) step design until obtain be satisfied with Control platform;

According to above-mentioned (1), the design tuning PID controller parameter of (2) and (3) step, this pid parameter setting method based on time scale, in (2) step, changes K _fbe equivalent to and change other 3 parameter K simultaneously _p, T _i, K _d, this is a kind of simple effective method, can obtain satisfied Control platform general 2 ~ 4 times.